Initiated in response to informal reports of vocabulary gains from gamers at universities in New Zealand and the Netherlands, this qualitative study explored how English language learners autonomously learn vocabulary while playing massively multiplayer online role-playing games (MMORPGs). Using research processes inherent in Grounded Theory, data was collected from a criterion sample of six experienced English-as-a-second-language players of World of Warcraft. Data included semi-structured interviews, observations, elicited email texts and extant texts. Data was transcribed, coded, and refined through a cyclic process of constant comparative analysis until patterns emerged. The results revealed 15 vocabulary learning strategies, including noticing frequency of words, recognizing knowledge gaps, selecting words for attention, equating images and actions to words, giving and receiving explanations and feedback, observing players, using words to learn words, reading in-game information, and using Google. This taxonomy of 15 vocabulary learning strategies provides a framework for further research into vocabulary learning strategy taxonomies and autonomous learning in informal digital contexts.